Information Fusion of Digital Camera and Radar

Jiexin Ren,Yao Wang,Yubing Han,Renli Zhang
DOI: https://doi.org/10.1109/imbioc.2019.8777799
2019-01-01
Abstract:It is found that the current traffic surveillance system is incapable of measuring multiple vehicles speed simultaneously. In this paper, a method that fuses the information of digital camera and radar has been proposed to improve the situation. First, You Only Look Once (YOLOv3) convolutional neural network is used to detect the vehicle, following with the Kalman filter to realize tracking. Then, the targets speed can be obtained from the radar through a series of processing, including mixing, sampling, filtering and signal processing. Finally, the camera calibration algorithm is utilized to combine the radar targets and video vehicle detection targets. It has been demonstrated that the speed of each vehicle can be accurately showed on the display.
What problem does this paper attempt to address?